The autistic spectrum: subgroups, boundaries, and treatment.

There is consensus about the disorders that comprise the autistic spectrum, with autistic disorder, Asperger's disorder, and PDD-NOS as the most typical examples and Rett's disorder and disintegrative disorder as the other components. Important controversies regarding the precise definitions of autistic spectrum disorders and the boundaries between the milder manifestations of those disorders, particularly PDD-NOS, and non-autistic conditions have not been and cannot be resolved fully as long as there is no known biologic cause or consistent biologic or psychological marker. This includes issues as basic as whether the autistic spectrum is a predominantly unitary entity or a collection of more or less similar phenotypes with multiple, varying etiologies. This is why the highest long-term priority in the area of definite diagnosis is the search for biologic marker(s) for autism and related autism spectrum disorders [91]. In the absence of a medical test to unequivocally diagnose autism, definitions of autism and related conditions are based only on manifestations in overt behavior, with all the unreliability this entails. In the future, the discovery of biologic correlates, causes, and pathogenetic pathways will undoubtedly change the way in which autism is diagnosed and lead to a new nosology [95]. Until that time the definitions in the current versions of the classification systems should be considered in a state of evolution. The key problem of the current classification systems is the fact that the boundaries between the various disorders are fuzzy. Instead of a categorical approach, a more useful description might be that of "autistic spectrum disorder," which reflects the range of severity of symptoms. Such a dimensional understanding of PDD is useful to clinicians, who may otherwise use nonspecific terms to avoid the categorical diagnosis of autism [31]. Rutter and Schopler [96] argued for separate clinical and research schemes because clinical and research needs are different. For research purposes it is desirable to have as much direct comparability across studies as possible. The focus is on a high degree of homogeneity within diagnostic groupings. A price must be paid for this detailed specification, and the main cost lies in the proportion of cases left undiagnosed. For example, there may be good scientific reasons for a narrowly defined categorical diagnosis that includes only individuals who definitely and clearly have a specifically defined condition and excludes individuals who may have the condition. For clinicians and educators, classification helps guide the selection of treatments for an individual. From this point of view, broader diagnostic concepts may be most appropriate [95]